Dynamics of Shoreline Changes on Bhasan Char Island, Bangladesh: A Comprehensive Analysis from 2005 to 2023 Utilizing DSAS and GIS
Shoreline delineation is crucial for understanding coastal dynamics and managing coastal resources effectively. Despite its importance, there is a lack of comprehensive research, especially in regions such as Bhasan Char Island, Bangladesh, where dynamic morphological changes pose challenges to sustainable management. The aim of the study is to provide a detailed analysis of shoreline changes on Bhasan Char Island, focusing on erosion and accretion trends across its different segments from 2005 to 2023, utilizing the Digital Shoreline Analysis System (DSAS) and Geographic Information System (GIS). The findings of the shoreline variation are presented in four distinct sections. The rate of shoreline change in the form of erosion and accretion is automatically assessed by four statistical parameters: Endpoint Rate (EPR), Net Shoreline Movement (NSM), Linear Regression Rate (LRR), and Shoreline Change Envelope (SCE). The results showed that the southwestern segment of Bhasan Char Island has experienced substantial erosion with an average Net Shoreline Movement (NSM) of \(-185.01 \, \text{m/year}\). However, the southeastern part of the char has shown significant accretion. The study used the EPR method to quantify shoreline movement, revealing accretion characteristics in approximately \(67.54\%\) of all transects with a mean accretion rate of \(157.63 \, \text{m/year}\). In contrast, erosion was observed in \(15.78\%\) of all transects with a mean erosion rate of \(-34.81 \, \text{m/year}\). The result of the current study provides crucial insights into the dynamic coastal processes of Bhasan Char, emphasizing the necessity for informed coastal management and planning.
1 Introduction
Shoreline refers to the physical boundary between land and water that provides valuable information about the coastal process and the dynamics of the shoreline (Dolan, Hayden, May, and May, 1980). They can also be interpreted as the transition zone between the marine and terrestrial environments (Oyedotun, 2014). Monitoring shorelines for the purpose of environmental protection is crucial (Rasuly, Naghdifar, and Rasoli, 2010). The alteration of shorelines is a significant concern in deltaic regions across the world (Chand and Acharya, 2010; Mills, Buckley, Mitchell, Clarke, and Edwards, 2005; Morton, 1979; Nandi, Ghosh, Kundu, Dutta, and Baksi, 2016; Rasuly et al., 2010; Ryabchuk, Spiridonov, Zhamoida, Nesterova, and Sergeev, 2012). Bangladesh is known as a low-lying deltaic country (Shibly and Takewaka, 2012), which stands among the world’s most susceptible countries to climate impacts due to the effects of global warming and rising sea levels (Brammer, 2014; Choudhury, Haque, and Quadir, 1997; Islam, Hossain, Hasan, and Murshed, 2016; Khan, 2013; Singh, 2002). A report conducted by UN scientists predicted that by 2050, around 17% of Bangladesh will be submerged, potentially resulting in about 30 million people becoming homeless (Dummett, 2008). Severe flooding during 1987, 1988, and 1996 drastically altered the country, transforming 60% of it into what appeared as temporary inland islands (Werle, Martin, and Hasan, 2000). The murky waters of the Ganges-Brahmaputra-Meghna River system carry sediments, gradually forming delicate landmasses known as Char islands (Banerjee, 2020a). These Char islands are consistently vulnerable to erosion and flooding. Furthermore, the Bay of Bengal is a precarious region where approximately 40% of annual global storm surges are documented in Bangladesh (Reduced Death Rates from Cyclones in Bangladesh, 2011). Some Char islands succumb to erosion during strong tides, while others stabilize over decades, serving purposes such as fishing and farming before eventually becoming inhabited.
Bhasan Char, meaning ’floating island,’ is located at the mouth of the Meghna River in southeast Bangladesh. The island undergoes dynamic changes due to the deposition of millions of tons of sediment from the Ganges and Jamuna rivers interacting with massive sea waves (Sarwar and Woodroffe, 2013). Emerged in 2006, approximately 30 km away from the mainland, the island is susceptible to severe erosion and accretion processes due to its spatial location (Bremner, 2020). Hence, conducting a reliable study on shoreline change rates is essential for strategic planning. Shorelines are extremely dynamic and complex due to various factors, including marine forcing such as waves, tides, and littoral drift; hydrodynamic forces like sea level rise, sediment dynamics, and water discharge; meteorological forcing such as severe storms, wind, and heavy precipitation; geophysical forcing such as sudden seismic events, and the formation of spits, barrier islands, and deltas; and anthropogenic forcing like urbanization and industrialization. The absence of clear definitions about the shoreline position creates challenges for undertaking statistical analysis of shorelines for hazard mapping and vulnerability zoning (Al Bakri, 1996; Zuzek, Nairn, and Thieme, 2003). The dynamic nature of coasts, shaped by tides, waves, and human activity, makes it challenging to analyze changes in the shoreline. Utilizing complex techniques and having knowledge of multiple scientific fields are necessary for gathering and analyzing data.
Traditionally, estimating the shoreline position has relied on field surveys, line tracing from toposheets, and aerial photos. Aerial photography, extensively utilized in the 1970s, introduced an error of 10 meters. Additionally, physical field surveys for shoreline analysis resulted in errors ranging from 3 to 4 meters (Morton, 1979). Currently, the utilization of remote sensing data, tools, and techniques in the analysis of historical shoreline changes has progressed significantly, with ongoing efforts to minimize associated uncertainties (Moussaid, For a, Zourarah, and Maanan, 2015). The Landsat imageries (TM, ETM+, and OLI) are widely employed for quantification in shoreline analysis globally (Bheeroo, Chandrasekar, Kaliraj, and Magesh, 2016; Murali et al., 2015; Nandi et al., 2016).
The DSAS is an add-in for Environmental System Research Institute’s (ESRI) ArcGIS that measures statistics related to shoreline changes using optical satellite imagery taken at different points in time (Thieler, Himmelstoss, Zichichi, and Ergul, 2009). The automated tool of DSAS developed by United States Geological Survey (USGS) has revolutionized both short- and long-term shoreline change analysis all over the world (Brooks and Spencer, 2010). The combination of ESRI ArcGIS and the DSAS tool facilitates shoreline extraction, baseline creation, transect generation, and computation of distances between the baseline and each transect using a user-friendly interface (Acharyya, Mishra, and Kar, 2020; Mishra et al., 2019; Nassar et al., 2019; Thieler et al., 2009). Researchers all over the world use the DSAS versions 1 through 5 as useful tools for monitoring shoreline positions and geometry, analysis of shoreline change, and assessment.
The DSAS tool has become indispensable in the realm of coastal studies, providing a systematic and precise approach to understanding the evolving coastlines. Its significance in facilitating the analysis of shoreline movements is well acknowledged in the literature. Himmelstoss et al. (E. A. Himmelstoss, 2009) emphasized the importance of DSAS in providing a comprehensive approach to quantify and analyze shoreline changes over time, showcasing its application in a variety of coastal settings. The tool’s ability to integrate with GIS for a detailed spatial analysis enhances its functionality, allowing for a more nuanced understanding of shoreline dynamics and informing coastal management practices. The advancements in remote sensing technology, coupled with the DSAS, have greatly expanded the possibilities for shoreline monitoring, providing researchers and practitioners with robust tools to assess, predict, and manage the impacts of shoreline changes.
As shorelines continue to change under the influence of both natural and anthropogenic factors, the role of technologies like DSAS in monitoring and predicting these changes becomes increasingly important. This study leverages these advancements, seeking to provide a comprehensive understanding of shoreline dynamics in Bhasan Char, which stands as a testament to the ever-evolving nature of deltaic environments. The insights gained from this study will contribute to the broader discourse on coastal management and resilience, aiding in the formulation of strategies to mitigate the impacts of shoreline changes in vulnerable regions.
Objectives
-
To analyze the shoreline change of the Bhasan Char Island from 2005–2023 using DSAS.
-
To analyze the coastal dynamics of Bhasan Char Island.
2 Literature Review
Understanding the dynamics of coastlines is crucial for effective coastal management. Shoreline change analysis plays a critical role in this project, providing insights into erosion and accretion patterns, informing coastal defense strategies, and assisting in the management of habitats and natural resources. This analysis has evolved significantly over time, transitioning from manual, error-prone methods to sophisticated, automated systems.
Traditionally, shoreline change analysis relied heavily on manual mapping techniques and aerial photographs. While this provided valuable information, it was susceptible to inconsistencies and inaccuracies due to observer bias, limited data resolution, and difficulties in comparing shorelines across different time periods. Additionally, early analyses primarily focused on short-term changes, failing to capture the broader trends and underlying geomorphological processes (Dolan et al., 1980).
A significant advancement in the field came with the development of the Digital Shoreline Analysis System (DSAS) by (Thieler et al., 2009). This ArcGIS extension revolutionized shoreline change analysis by enabling researchers to systematically quantify shoreline movement through statistically robust methods. DSAS offers multiple shoreline extraction techniques, calculates various rate-of-change statistics (e.g., End Point Rate, Average of Rates, Least Squares Regression), and facilitates data visualization and statistical analysis. The system’s standardized approach minimizes observer bias and improves reproducibility compared to manual methods, making it a valuable tool for both long-term and short-term change analysis.
Researchers have actively compared different shoreline change analysis methods to identify the most accurate and reliable for specific scenarios. Genz et al. (Genz, Fletcher, Dunn, Frazer, and Rooney, 2007) emphasized the importance of tailoring the approach to the study’s objectives and regional characteristics. Their work on the Hawaiian coast highlighted the limitations of certain methods in predicting future shoreline changes, urging researchers to consider factors such as wave climate, coastal morphology, and human modifications when selecting an analysis method. Meanwhile, Himmelstoss et al. (E. Himmelstoss, Henderson, Kratzmann, and Farris, 2018a) expanded DSAS’s capabilities and provided an updated guide reflecting ongoing improvements. They addressed limitations in applying DSAS to non-traditional coastal environments and incorporated advanced statistical techniques for improved accuracy.
Numerous case studies demonstrate the diverse applications of DSAS and other methods in different geographic locations. Moore (Moore, 2000) reviewed shoreline mapping techniques, emphasizing the evolution of mapping approaches and their crucial role in various coastal management scenarios. These studies exemplify the adaptability of shoreline change analysis tools to different data availabilities, study objectives, and coastal environments. They underscore the importance of accurate shoreline mapping not only for coastal management but also for understanding regional geological changes and ecological processes.
A prominent theme throughout the literature is the critical need to select the appropriate method for each shoreline change analysis. The optimal method depends on several factors, including the specific goals of the study (e.g., erosion risk assessment, habitat modeling), the nature of the shoreline (e.g., sandy beach, rocky cliff), available data (e.g., historical aerial photographs, LiDAR data), and the desired temporal and spatial scales of analysis. While tools like DSAS have transformed the field, researchers consistently emphasize the importance of understanding the inherent limitations and potential errors of each method. Critical evaluation of method selection ensures reliable and meaningful results that can inform effective coastal management strategies.
The field of shoreline change analysis has come a long way from its manual and error-prone beginnings. The development of tools like DSAS has revolutionized the field, but research continues to strive for more reliable, efficient, and applicable methods. As technology and methodology advance, research efforts focus on incorporating additional data sources (e.g., satellite imagery, bathymetry data), utilizing machine learning algorithms for automated shoreline extraction, and developing predictive models for long-term shoreline change under the influence of factors like sea-level rise and climate change. Ultimately, continuous refinement of shoreline change analysis techniques will lead to improved coastal management strategies, ensuring the sustainable future of our coastlines and the communities that depend on them.
3 Materials and Methods
3.1 Study area
Bhasan Char, also known as Thengar Char or Floating Island, is an uninhabited island in the Bay of Bengal. Geographically, the island is located within the jurisdiction of Noakhali District of Bangladesh, specifically in the Sandwip Upazila. It extends from \(22^\circ 20' \, \text{N}\) to \(22^\circ 27' \, \text{N}\) latitudes and from \(91^\circ 21' \, \text{E}\) to \(91^\circ 27' \, \text{E}\) longitudes. The island is surrounded by the Bamni River to the north, the Meghna River to the west, Sandwip Island to the northeast, Hatiya Island to the southwest, and the Bay of Bengal to the south. Bhasan Char is approximately \(24 \, \text{km}\) away from Hatiya and \(6 \, \text{km}\) from Sandwip Island. The total areal extent of the island is nearly \(80 \, \text{km}^2\), and the average elevation is about \(2.84 \, \text{m}\) above mean sea level (Banerjee, 2020b). On this island, monthly temperatures generally fluctuate between \(27^\circ \text{C}\) and \(36^\circ \text{C}\). January experiences the lowest rainfall, around \(2.18 \, \text{mm}\), while July has the highest, approximately \(333 \, \text{mm}\). Humidity levels vary from \(61\%\) to \(81\%\) throughout the year (Braun, Wagner, and Hochschild, 2018). The monthly average wind speed ranges from \(7.7 \, \text{km/h}\) in January to \(19.7 \, \text{km/h}\) in June, and the average monthly atmospheric pressure fluctuates between \(1000.6\) and \(1014 \, \text{millibars}\) (Bremner, 2020). The island is shaped by the Meghna estuarine system and is part of the dynamic Ganga Brahmaputra–Meghna deltaic region, where substantial sediment deposition occurs from the upper catchments. The island’s size undergoes considerable fluctuations due to ongoing accretion and erosion processes, ranging from \(22 \, \text{km}^2\) in 2005 to \(77.48 \, \text{km}^2\) in 2023. The island is still experiencing significant landmass changes and natural consolidation processes. To determine the rapid changes in shoreline and area since its evolution, the study area has been segmented into four parts (Segment A, B, C, and D) for better understanding and analysis.
3.2 Data Sources
The major purpose of using satellite images was to assess shoreline dynamics, measuring erosion and accretion with the annual rate. In this study, Multi-temporal Landsat satellite images (Landsat MSS, TM, ETM+, OLI-TIRS) from 2005-2023 were utilized with three-year interval to monitor shoreline changes on Bhasan Char. Landsat MSS, TM, ETM+, OLI-TIRS data used in this study were downloaded from the United States Geological Survey (USGS). The selection of images was based on criteria including less than 10% cloud cover, good visibility, and high quality.
| SENSOR PLATFORM | SENSOR | ACQUISITION DATE | SPATIAL RESOLUTION (m) | CLOUD COVER |
|---|---|---|---|---|
| Landsat 5 | TM | 07-Apr-2005 | \(30\) | cloud free |
| Landsat 7 | ETM+ | 06-Aug-2008 | \(30\) | \(3\%\) |
| Landsat 7 | ETM+ | 23-Aug-2011 | \(30\) | \(2\%\) |
| Landsat 7 | ETM+ | 18-Mar-2014 | \(30\) | cloud free |
| Landsat 7 | ETM+ | 22-Dec-2017 | \(30\) | \(1\%\) |
| Landsat 7 | ETM+ | 04-Sep-2022 | \(30\) | \(5\%\) |
| Landsat 8 | OLI | 23-Aug-2023 | \(30\) | \(2\%\) |
3.3 Software
In this study, I used Environmental Systems Research Institute’s (ESRI) ArcGIS 10.5 software for vector generation, analysis, editing, and map composition. Google Earth Engine was used for shoreline extraction and QGIS 3.32 to fix scan line errors in Landsat 7 ETM+ imageries. The United States Geological Survey’s (USGS) Digital Shoreline Analysis System (DSAS 5.0) was used for cast transect and shoreline change rate calculations.
3.4 Image Processing
Due to the malfunction of the Scan Line Corrector (SLC) in the Enhanced Thematic Mapper Plus (ETM+) sensor aboard Landsat 7, imagery acquired post-failure exhibits systematic, wedge-shaped data voids on both sides of each scene. This anomaly results in an estimated data loss of approximately 22%. The image processing carried out in this study involved strip filling to mitigate the scan line error. Gap filling was applied comprehensively to the Landsat-7 ETM+ imagery across all spectral bands utilizing QGIS 3.32. All the image were projected to World Geodetic System(WGS1984) datum and Universal Transverse Mercator (UTM) zone 46N projection. After georeferencing, (RMSE) was found to be less than 0.5 pixels, indicating that the images were geometrically well-matched.
3.5 Shoreline Extraction
Shorelines are the high water line as surveyed by GPS units in kinematic mode (Moore, 2000). Meanwhile, automatic coastline demarcation from low-resolution satellite images is a complicated job due to the unclear boundary between water and land in the saturated zone (Maiti and Bhattacharya, 2009). In this study, the shoreline was delineated from satellite imagery utilizing Google Earth Engine. The following flowchart outlines the methodology employed for the extraction of the shoreline:
In the initial phase of the analysis, I generated a cloud-free composite image to ensure the highest possible clarity and data integrity for the subsequent stages of waterbody extraction. I then employed a robust algorithm to delineate all waterbodies present within the satellite imagery. To refine the analysis to coastal areas, I systematically removed inland water features and small islands, which do not contribute to the definition of the shoreline. Following this, I converted the raster representation of waterbodies to a vector format, which allowed for greater manipulation and precision. Lastly, I applied a simplification algorithm to the vector data, meticulously extracting the coastline.
3.6 Digital Shoreline Analysis System (DSAS)
For the purpose of this study, I utilized the Digital Shoreline Analysis System (DSAS) Version 5.0, developed by the United States Geological Survey (USGS), to assess the rate of shoreline changes. DSAS 5.0 is a tool that functions as an add-on to the ArcGIS 10.5 software(Kale, Ataol, and Tekkanat, 2019). It computes statistical changes in multiple shoreline positions and has been widely employed in the literature. To assess the temporal and geometric changes in shoreline movement, a buffer analysis (500 m) was employed on the shoreline layers subsequent to their formation. This process resulted in the establishment of a hypothetical baseline. DSAS generates cross-sections (transects) perpendicular to the defined baseline. Subsequently, the positions of the shoreline along the baseline are recorded(E. A. Himmelstoss, 2009).Leatherman et al.(Leatherman, Zhang, and Douglas, 2000)suggested that computing the long-term shoreline change rates for the average beach behavior can be achieved through three methods. These methods include simple averaging of all shoreline rates, averaging shoreline change rates specifically for erosion and accretion areas(Aiello, Canora, Pasquariello, and Spilotro, 2013), and averaging shoreline change only for the erosion sections unaffected by inlets and coastal engineering.
DSAS offers five statistical approaches for calculating shoreline change, including Shoreline Change Envelope (SCE), Net Shoreline Movement (NSM), Weighted Linear Regression (WLR), the End Point Rate (EPR), and Linear Regression Rate (LRR)(Thieler et al., 2009). Linear regression is employed in the long-term shoreline change estimation process(Crowell, Douglas, and Leatherman, 1997). The calculation of LRR, EPR, and WLR requires three or more shorelines, while NSM and SCE use two shoreline positions for the calculation. The rate of changes is expressed in meters per year (m/year), whereas the distance is expressed in meters.
a) The NSM method was utilized to calculate the distance between the oldest and most recent shorelines for each transect, employing only two historical shoreline positions. The formula for NSM is illustrated as follows(Selamat et al., 2019)
where \(S_m\) denotes the net
shoreline movement (m), \(f_n\) denotes
the distance between the baseline and shoreline for the oldest shoreline
(m), and \(f_m\) denotes the distance
between the baseline and shoreline for the most recent shoreline
position along the same transect. In fact, both NSM and SCE do not
report the rate value, but these methods represent the shoreline changes
in distance value.
b) Shoreline Change Envelope (SCE) represents the maximum distance
between the oldest and youngest shorelines, measured in meters. The
equation below was used to calculate SCE(Selamat et
al.,
2019):
where \(S_d\) represents the shoreline distance (m); \(f_x\) is the distance of the further shoreline from the baseline (m); and \(f_y\) is the distance of the shoreline closest to the baseline (m).
c) In the EPR method, the calculation is performed by dividing the distance between the shoreline at the oldest and the most recent dates by the time elapsed(To and Thao, 2008). It is the ratio of NSM to that of time elapsed (between the oldest and the youngest shoreline). The equation for the calculated statistics is:
d) Linear Regression Rate (LRR) is the slope of the line, which is determined by fitting a least square regression line to all transects intersecting the shorelines.The equation for the calculated linear regression line is:
e) Weighed Linear Regression (WLR): As compared to the LRR, it gives more emphasis or weight to the reliable data to determine the best-fit line in WLR (Genz et al., 2007). A position with less uncertainty has high emphasis in this method. It is calculated as:
where \(w\) is weight, and \(e\) is the
shoreline uncertainty
value.
In this study, three specific statistical operations were employed, namely, Shoreline Change Envelope (SCE), Net Shoreline Movement (NSM), and End Point Rate (EPR). The benefit of using these statistical operations is the ability to compute the rate-of-change statistics for a time series of shoreline positions. The statistics allow the nature of shoreline dynamics and trends in change to be evaluated and addressed.
4 Results and Discussion
4.1 Restult of DSAS
Transects are straight lines originating from the baseline, evenly spaced among themselves. They intersect with the shorelines considered for the study and are employed to measure changes observed in the shoreline.
The curvilinear shorelines are segmented into four distinct sections—Northwestern (A), Northeastern (B), Southwestern (C), and Southeastern (D)—with a total of 103 transect lines distributed among these segments. The distribution includes 12 transect lines in segment A, 28 in segment B, 46 in segment C, and 28 in segment D. These transects are spaced at regular intervals with a 70-meter gap between them, collectively covering a shoreline distance of 39,140 meters (39.14 km). NSM and SCE are utilized to determine variations in shoreline position from the year 2005 to 2023. The Jenks Natural Breaks method is applied for transect classification in all cases (NSM, SCE, EPR, LRR) for shoreline change calculations for the period between 2005 and 2023.
4.1.1 Net Shorelie Movement(NSM)
| Section | Transects Length (m) | Transect Count | Transect Count (%) | Movement | Avg. NSM |
|---|---|---|---|---|---|
| A | \(1484.70\) – \(1719.06\) | \(3\) | \(25.000\) | Seaward | \(1669.81\) |
| \(1719.06\) – \(2067.55\) | \(1\) | \(8.333\) | Seaward | \(2067.55\) | |
| \(2067.55\) – \(2565.01\) | \(1\) | \(8.333\) | Seaward | \(2565.01\) | |
| \(2565.01\) – \(2665.64\) | \(1\) | \(8.333\) | Seaward | \(2665.64\) | |
| \(2665.64\) – \(2783.69\) | \(3\) | \(25.000\) | Seaward | \(2768.84\) | |
| \(2783.69\) – \(2833.97\) | \(2\) | \(16.667\) | Seaward | \(2828.03\) | |
| B | \(-339.40\) – \(-49.58\) | \(4\) | \(12.500\) | Landward | \(-60.77\) |
| \(-49.58\) – \(1738.50\) | \(5\) | \(15.625\) | Landward | \(1612.2\) | |
| \(1738.50\) – \(2632.50\) | \(1\) | \(3.125\) | Seaward | \(2445.53\) | |
| \(2632.50\) – \(2981.50\) | \(2\) | \(6.250\) | Seaward | \(2896.92\) | |
| \(2981.50\) – \(3473.50\) | \(4\) | \(12.500\) | Seaward | \(3345.25\) | |
| \(3473.50\) – \(3808.50\) | \(7\) | \(21.875\) | Seaward | \(3704.94\) | |
| \(3808.50\) – \(4152\) | \(4\) | \(12.500\) | Seaward | \(3984.37\) | |
| C | \(-723.22\) – \(-142.4\) | \(6\) | \(50.00\) | Landward | \(-185.01\) |
| \(-142.45\) – \(301.68\) | \(1\) | \(8.33\) | Landward | \(301.68\) | |
| \(301.68\) – \(1892.85\) | \(1\) | \(8.33\) | Seaward | \(1892.85\) | |
| \(1892.85\) – \(2694.11\) | \(1\) | \(8.33\) | Seaward | \(2694.11\) | |
| \(2694.11\) – \(3193.37\) | \(1\) | \(8.33\) | Seaward | \(3193.37\) | |
| \(3193.37\) – \(3512.88\) | \(1\) | \(8.33\) | Seaward | \(3512.88\) | |
| D | \(1687.50\) – \(1763.31\) | \(4\) | \(14.81\) | Seaward | \(1740.11\) |
| \(1763.31\) – \(1929.77\) | \(3\) | \(11.11\) | Seaward | \(1870.17\) | |
| \(1929.77\) – \(2054.89\) | \(8\) | \(29.63\) | Seaward | \(2028.71\) | |
| \(2054.89\) – \(2208.04\) | \(4\) | \(14.81\) | Seaward | \(2176.14\) | |
| \(2208.04\) – \(3274.95\) | \(4\) | \(14.81\) | Seaward | \(3082.88\) | |
| \(3274.95\) – \(3668.74\) | \(3\) | \(11.11\) | Seaward | \(3561.96\) | |
In Section A , the table (Table 2) provides details on transect lengths, including their frequency, percentage, directional movement (either seaward or landward gain), and average Net Shoreline Movement (NSM). Each transect length, measured in meters, represents the observed range of shoreline change within the section. The table indicates that all transects in Section A exhibit seaward movement, signifying a net gain in the seaward direction. No instances of landward movement are observed in this section. The maximum recorded seaward movement is 2768.84 meters. On average, there is a seaward gain of approximately 2427.48 meters, representing the mean of the average distances of shoreline advancement seaward across the specified ranges in Section A.
In contrast to Section A, Section B demonstrates both landward and seaward movements, reflecting a more dynamic shoreline change scenario. The table (Table 2) delineates these changes across different transect lengths in meters, representing the range of shoreline change observed. The transect lengths vary, with the first two ranges showing negative average NSM values, indicating landward movement or erosion. The subsequent ranges all show positive average NSM values, signifying seaward gains.The highest seaward movement is observed in the transect length range of 3809 to 4152 meters, with an average NSM of 3984.365 meters.
Section C of the NSM record(Table 2) showcases a combination of landward and seaward movements. The transect range of -723.22 to -142.45 meters shows the most substantial landward movement with an average NSM of -185.01 meters, suggesting erosion in this area. On the other hand, the 3193.37 to 3512.88 meter range exhibits the highest seaward movement with an average NSM of 3512.88 M, reflecting significant accretion in this section.
Section D of the NSM record(Table2) shows an exclusive pattern of seaward movement across all transect length ranges, indicating a consistent trend of shoreline accretion in this segment of the study area. The highest seaward movement is 3561.96 meter indicating the most significant average shoreline advancement seaward in this section.
4.1.2 Shoreline Change Envelope(SCE)
The computation of the Average Shoreline Change Envelope (SCE) involves summing all recorded SCE values within a specified range and dividing the sum by the number of transects within that range. The SCE outcomes provide insights into the shoreline extent, highlighting the distance between the furthest and closest points to the baseline along the transects. By calculating the average SCE for each range, the study discerns noteworthy variations in shoreline movement among the four distinct sections, emphasizing the dynamic nature of the coastal environment.
| Section | Transect Length (m) | Transect Count | Transect Count (%) | Average SCE |
|---|---|---|---|---|
| A | \(3051.06\) – \(3132.21\) | \(1\) | \(8.33\) | \(3132.21\) |
| \(3132.21\) – \(3217.77\) | \(2\) | \(16.67\) | \(3206.44\) | |
| \(3217.77\) – \(4321.75\) | \(2\) | \(16.67\) | \(4249.15\) | |
| \(4321.75\) – \(4610.97\) | \(2\) | \(16.67\) | \(4540.61\) | |
| \(4610.97\) – \(4972.01\) | \(2\) | \(16.67\) | \(4859.59\) | |
| \(4972.01\) – \(5336.27\) | \(2\) | \(16.67\) | \(5269.08\) | |
| B | \(47.75\) – \(86.07\) | \(7\) | \(19.44\) | \(70.46\) |
| \(86.07\) – \(1346.00\) | \(0\) | \(0.00\) | \(0.00\) | |
| \(1346.00\) – \(1485.00\) | \(3\) | \(8.33\) | \(1484.42\) | |
| \(1485.00\) – \(3252.00\) | \(9\) | \(25.00\) | \(2920.44\) | |
| \(3252.00\) – \(4265.00\) | \(2\) | \(5.56\) | \(4075.34\) | |
| \(4265.00\) – \(4708.00\) | \(3\) | \(8.33\) | \(4543.45\) | |
| C | \(142.45\) – \(228.23\) | \(3\) | \(7.14\) | \(196.36\) |
| \(228.23\) – \(1655.38\) | \(5\) | \(11.90\) | \(1436.36\) | |
| \(1655.38\) – \(2866.49\) | \(5\) | \(11.90\) | \(2655.43\) | |
| \(2866.49\) – \(3378.79\) | \(3\) | \(7.14\) | \(3211.36\) | |
| \(3378.79\) – \(3915.75\) | \(4\) | \(9.52\) | \(3730.73\) | |
| \(3915.75\) – \(4333.31\) | \(6\) | \(14.29\) | \(4151.14\) | |
| D | \(1934.19\) – \(2242.40\) | \(4\) | \(14.29\) | \(2059.78\) |
| \(2242.40\) – \(3395.16\) | \(6\) | \(21.43\) | \(3212.27\) | |
| \(3395.16\) – \(3959.59\) | \(4\) | \(14.29\) | \(3728.13\) | |
| \(3959.59\) – \(4412.99\) | \(5\) | \(17.86\) | \(4259.99\) | |
| \(4412.99\) – \(4933.12\) | \(5\) | \(17.86\) | \(4707.92\) | |
| \(4933.12\) – \(5769.64\) | \(2\) | \(7.14\) | \(5577.84\) | |
According to the results3, section C was identified with the minimum SCE value at 47.75 m4. Meanwhile, section D recorded the highest SCE value at 5769.64 m. Section A shows the least variability, with all transects (100%) indicating significant shoreline shifts and a high average SCE of 4202.75 meters, suggesting a uniformly dynamic shoreline. Section D records the highest maximum SCE at 5769.64 meters, indicating extreme changes in certain areas, with nearly all transects (96.43%) showing significant shifts. Sections B and C exhibit more moderate dynamics, with a lower average SCE and fewer significant shifts, indicating less dramatic changes in the shoreline. Section B exhibits the smallest average SCE at 2189.4 meters among all sections, with 52.78% of its transects demonstrating notable shifts. This suggests that while the changes are less pronounced, there is still a significant level of activity. Conversely, Section C presents a lower average SCE, underscored by a substantial count of transects (15 out of 42) recording zero SCE, indicating a predominantly stable area with less variability in shoreline movement.
| Section A | Section B | Section C | Section D | |
|---|---|---|---|---|
| SCE (max) | \(5336.27\) | \(4708.10\) | \(4333.31\) | \(5769.64\) |
| SCE (min) | \(3051.06\) | \(47.75\) | \(142.45\) | \(1934.19\) |
| SCE (Avg) | \(4202.75\) | \(2189.40\) | \(2616.83\) | \(3716.84\) |
4.1.3 Long-term shoreline changes(EPR Model)
As all data are obtained from the same source, Google Earth Engine(Google Earth Engine, n.d.), a uniform ‘uncertainty’ value (DSAS_uncy) of 1 is applied to all data points. In weighted linear regression, data with lower uncertainty are given greater emphasis, thereby significantly influencing the determination of the best-fit line. This approach is particularly pertinent in the computation of rate-of-change statistics for shorelines, where data points with lesser position uncertainty are weighted more heavily.
The weight (\(w\)) is defined as a function of the variance in the measurement’s uncertainty (Equation 5)(E. Himmelstoss, Henderson, Kratzmann, and Farris, 2018b), ensuring that the most reliable data exerts a more substantial impact on the statistical analysis. In this study, LRR and WLR appeared the same due to uniform (DSAS_uncy). Hence, the \(R^2\) values indicate that the End Point Rate (EPR) was the most appropriate metric for explaining the long-term changes observed in the Bhasan Char shoreline.
The EPR is simply calculated by dividing the distance (m) separating two shorelines by the number of years between the dates of the two shorelines(Equation:3).This method is enormously prevalent in shoreline movement rate calculations, as it is widely used by different coastal researchers(Ayadi, Boutiba, Sabatier, and Guettouche, 2016; Dolan, Fenster, and Holme, 1991; Hwang, Choi, and Choi, 2014; Thieler et al., 2009).
The section-wise distribution produced by the long-term shoreline analysis of Bhasan Char shoreline is shown in Figure 18.
| Descriptive statistics (Based on EPR) | Section A | Section B | Section C | Section D |
|---|---|---|---|---|
| Total number of transects | \(12\) | \(28\) | \(46\) | \(28\) |
| Total number exhibiting erosion | \(0\) | \(10\) | \(8\) | \(0\) |
| Total number exhibiting accretion | \(12\) | \(17\) | \(21\) | \(27\) |
| Total number exhibiting stability | \(0\) | \(1\) | \(17\) | \(1\) |
| Percentage exhibiting erosion | \(0.0\) | \(35.71\) | \(17.39\) | \(0.0\) |
| Percentage exhibiting accretion | \(100.0\) | \(60.71\) | \(45.65\) | \(96.43\) |
| Percentage exhibiting stability | \(0.0\) | \(3.57\) | \(36.96\) | \(3.57\) |
| Mean shoreline change (m/year) | \(125.70\) | \(103.00\) | \(133.79\) | \(120.79\) |
| Maximum shoreline change (m/year) | \(154.07\) | \(401.84\) | \(798.01\) | \(199.45\) |
| Minimum shoreline change (m/year) | \(80.72\) | \(-28.28\) | \(-74.98\) | \(0.0\) |
| Standard deviation (m/year) | \(29.96\) | \(116.04\) | \(207.54\) | \(40.94\) |
| Mean erosion rate (m/year) | \(0\) | \(-22.72\) | \(-46.90\) | \(0\) |
| Standard deviation of erosion rate (m/year) | \(0\) | \(4.18\) | \(25.81\) | \(0\) |
| Mean accretion rate (m/year) | \(125.70\) | \(176.95\) | \(202.63\) | \(125.26\) |
| Standard deviation of accretion rate (m/year) | \(29.96\) | \(78.65\) | \(205.03\) | \(34.04\) |
The mean shoreline change rate, the maximum rate of change, the minimum rate of change, progress, regression, and the percentage of erosion and accretion have been compiled and are listed in(Table5. According to the End Point Rate (EPR) method, 77 out of 114 transects had accretion characteristics, and 18 transects showed erosion trends. The remaining 19 (16.66%) transects represented the stable coast in the shoreline change analysis.
Accretion was recorded on approximately \(67.54\%\) of all transects, with a mean accretion rate of \(157.63 \, \text{m/year}\). However, only \(15.78\%\) of all transects of the Bhasan Char coastline had erosion, with a mean erosion rate of \(-34.81 \, \text{m/year}\).
4.2 Dynamics of Shoreline
The initial length of the shoreline in 2005 recorded approximately 23.34 kilometers. Over the subsequent years, the shoreline exhibited a dynamic trend, experiencing fluctuations before steadily expanding to 35.28 kilometers by the year 2023, as detailed in (Table6) 6. Notably, the most substantial increase within a single period was observed between 2008 and 2011, indicating a remarkable gain of 9.29 kilometers. The cumulative extension of the shoreline from 2005 to 2023 reveals an overall growth of 11.94 kilometers, underscoring the significant changes in the coastal landscape over the studied timeframe.
| Year | Shoreline Length (km) | Shoreline Gain/Loss (km) |
|---|---|---|
| \(2023\) | \(35.28\) | \(2.73\) |
| \(2020\) | \(32.55\) | \(-3.08\) |
| \(2017\) | \(35.63\) | \(-3.51\) |
| \(2014\) | \(39.14\) | \(1.98\) |
| \(2011\) | \(37.16\) | \(9.29\) |
| \(2008\) | \(27.87\) | \(4.53\) |
| \(2005\) | \(23.34\) | - |
| 2005–2023 | \(11.94\) | |
4.3 Erosion and Accretion
Erosion and accretion are common in coastal areas, as they are dynamic in nature. Bhashan Char, located approximately \(39\, \text{km}\) away from Noakhali in an estuary of the Meghna River, is particularly dynamic due to the large volume of sediment deposited by the Ganges-Brahmaputra-Meghna (GBM) river systems. From \(2005\) to \(2023\), Bhasan Char experienced significant changes in land area due to erosion and accretion. The study revealed that erosion processes were more severe than accretion, resulting in a loss of approximately \(231.34\, \text{km}^2\) of land. The period from \(2014\) to \(2017\) witnessed the highest erosion rate at \(27.17\, \text{km}^2/\text{year}\), indicating significant land loss. Concurrently, accretion peaked during the same period at \(32.21\, \text{km}^2/\text{year}\), reflecting substantial land gain. The lowest erosion rate occurred during the period \(2008-11\), at \(0.9467\, \text{km}^2/\text{year}\), indicating a temporary period of relative stability. Notably, there was an increase in the erosion rate from \(2008-11\) to \(2011-14\). The data reflect a delicate balance with fluctuations in erosion and accretion rates, emphasizing the dynamic nature of the island’s geography.
| Duration | Erosion (\(\mathrm{km}^2\)) | Accretion (\(\mathrm{km}^2\)) | Common Land (\(\mathrm{km}^2\)) |
|---|---|---|---|
| 2005–08 | \(11.59\) | \(19.49\) | \(11.23\) |
| 2008–11 | \(2.84\) | \(19.87\) | \(27.88\) |
| 2011–14 | \(46.71\) | \(14.34\) | \(43.09\) |
| 2014–17 | \(81.51\) | \(96.6246\) | \(46.47\) |
| 2017–20 | \(76.08\) | \(14.63\) | \(51.64\) |
| 2020–23 | \(12.61\) | \(12.450651\) | \(65.012761\) |
The changing pattern of the island is not uniform. Shapefiles from different periods are utilized to estimate the erosion and sedimentation occurring on Bhasan Char. To determine the extent of erosion, the landmass of the study area in a subsequent year must be subtracted from that of the previous year. Conversely, to identify the accreted area, the landmass of a given year must be subtracted from that of the subsequent year. The maps generated (Fig:12) from these procedures illustrate the areas where erosion and accretion are occurring. This may help in understanding the dynamics of landform changes and in detecting areas that require management.
The bar chart reveals the trends of Erosion, Accretion, and Common Land areas from 2005 to 2023, each exhibiting distinctive patterns of change over time.
Erosion shows a sharp rise from 2005 to 2011, where it reaches its peak in the 2011-14 period, indicating a period of intense land degradation. Following this peak, Erosion enters a significant downward trend, suggesting a substantial reduction in land being lost to natural or human-induced factors. Accretion, representing land gains, initially displays a downward trend until 2011, hinting at a slower rate of land formation or recovery. This trend reverses from 2011 to 2014, where Accretion sees a steep upward movement, peaking in 2014-17, implying an accelerated rate of land gain during this time. However, this is followed by a decrease towards the end of the period, indicating a reduction in the rate of new land formation or recovery. Common Land, which likely represents areas not affected by Erosion or Accretion, shows the most stability, yet still undergoes fluctuations. It experiences a slight initial increase, a notable decrease until 2014, and then a substantial rise to its highest point in 2014-17. The final years show a slight decline, but overall, the Common Land area remains relatively stable compared to the other two categories.
5 Conclusion
This study offers a comprehensive analysis of the shoreline changes in Bhasan Char, revealing a dynamic interplay of erosion and accretion processes across its four sections. Time series analysis of Landsat imagery from 2005 to 2023 have revealed that Section C(Southwestern segment) of the Bhasan Char Island experienced the most substantial landward movement, with an average Net Shoreline Movement (NSM) of -185.01 meters. This range of -723.22 to -142.45 meters indicates severe erosion in that specific area of the shoreline whereas section D(Southeastern segment) shows an exclusive pattern of seaward movement across all transect length ranges, indicating a consistent trend of shoreline accretion in this part of the study area. The most significant average shoreline advancement seaward in this section is 3561.96 meters. The study effectively utilized the End Point Rate (EPR) method to quantify shoreline movement, finding accretion characteristics in approximately 67.54% of all transects with a mean accretion rate of 157.63 m/year. Conversely, erosion was observed in 15.78% of all transects, with a mean erosion rate of -34.81 m/year.
A notable aspect of this study was the analysis of Long-term Rate of Change (LRR) and Weighted Linear Regression (WLR), which both appeared the same due to the uniform ’uncertainty’ value applied to all data points sourced from Google Earth Engine. This uniformity led to a greater emphasis on data with lower uncertainty in the weighted linear regression, significantly influencing the determination of the best-fit line. Consequently, the similar results between LRR and WLR underscore the importance of considering data uncertainty in shoreline change analysis.The study’s findings suggest that the consistent application of a uniform ’uncertainty’ value across all data points significantly influenced the outcome. It underscores the need for precise, accurate data for shoreline change analysis.